Micro–adhesion rings surrounding TCR microclusters are essential for T cell activation

To analyze the contribution of the LFA-1–mediated outside-in signal in the IS formation and T cell activation, we analyzed LFA-1–related molecules using a planar bilayer containing moth cytochrome C (MCC)–loaded MHC class II I-E^k and ICAM-1. T cells from AND-TCR transgenic (tg) mice were transduced with GFP-labeled molecules and stained with Dylight549 (Dy549)-labeled anti–TCR-β (H57) Fab to visualize the TCR. The images of fluorescence at the interface between the T cell and the bilayer were collected by total internal reflection fluorescence (TIRF) microscopy.

Staining of LFA-1 and TCR on fixed T cells revealed that LFA-1 formed small rings surrounding TCR-MCs (Fig. 1 A). GFP-labeled Pxn (Pxn-GFP) in live T cells showed clearer rings surrounding TCR-MCs (Fig. 1 B, left). The fluorescent intensity profiles of Pxn-GFP and TCRs showed that the fluorescent peaks did not overlap and distributed in an alternating way (Fig. 1 B, middle). The nonoverlapping feature was confirmed by plotting normalized pixel intensities of Pxn-GFP versus TCRs showing low correlation (Fig. 1 B and Fig. 2 C, right). The distribution and dynamics of the rings of Halo tag–labeled Pxn (Pxn-Halo) were almost identical to focal adhesion kinase Pyk2-GFP (Fig. 1 C). The fluorescent intensity profiles of Pyk2-GFP and Pxn-Halo showed their colocalization (Fig. 1 C, middle). The plot of normalized pixel intensities of Pyk2-GFP versus Pxn-Halo revealed a highly positive correlation (Figs. 1 C and 2 C, right). Staining of vinculin on fixed T cells as an endogenous focal adhesion molecule also showed similar rings (Fig. 1 D). Therefore, both integrin and focal adhesion molecules formed the micro–adhesion ring surrounding TCR-MCs. The criteria of the micro–adhesion ring are shown in Fig. 1 F. The concentric arrangement of TCRs and micro–adhesion rings resembles mature ISs in microscale.

Analysis of sequential images of TCR and Pyk2-GFP at the adhesion area revealed that the formation of micro–adhesion rings is transient (Fig. 1 E and Video 1). The generation of micro–adhesion rings began just after the initial TCR-MC formation, and their numbers increased with increasing TCR-MCs. However, these structures disappeared before cSMAC formation. The images of single micro–adhesion rings chronologically depicted and an array illustrating the duration of individual TCR-MCs and micro–adhesion rings showed that the micro–adhesion ring was formed just after TCR-MC generation at a very early phase of activation but soon disappeared (Fig. 1 G).

We confirmed these synapse-like structures in microscale at the interface of T cells and APCs (Fig. 2). T cells expressing SLP76-GFP were incubated with Ag-pulsed B cells and stained with anti–LFA-1 antibody (Ab). The­ x–y dimensional analysis at the interface revealed complex distributions of SLP76-GFP and LFA-1 (Fig. 2, A and B). In SLP76-GFP–positive cells conjugated with APCs, 45.2 ± 18.2% of the cells showed SLP76-GFP clusters and LFA-1 staining ordered alternately at the interface. The T cells expressing SLP76-GFP without clustering would have ceased the formation of SLP76 clusters because the SLP76 cluster is generated transiently at the very early activation stage (see Fig. 3 G). The number of SLP76 clusters between LFA-1 staining was evaluated. From the fluorescent intensity profiles shown in Fig. 2 C (left), the number of SLP76 clusters was estimated as 4.8 ± 2.3 on a diameter and 18.6 ± 9.18 per contact area. The plot of normalized pixel intensities of SLP76-GFP versus LFA-1 from representative cells (middle) was also shown. The means of correlation coefficients of SLP-GFP and LFA-1 from 10 cells were plotted in Fig. 2 C (right). These data clearly showed that SLP76 and LFA-1 were nonoverlapped. As a colocalization positive control, the higher correlation coefficient of Pyk2-GFP and Pxn-Halo was plotted (Fig. 1 C). The lower correlation coefficient of TCR and Pxn-GFP was also plotted for confirmation of similar value from different experimental conditions (Fig. 1 B). We also reconstituted the x–z dimension image and confirmed that SLP76-GFP clusters were surrounded by LFA-1 (Fig. 2 D). These data clearly show that the TCR-MCs in the early phase of activation induce the transient formation of synapse-like structures with micro–adhesion rings.

For quantitative analysis, the number of micro–adhesion rings was counted manually using the criterion explained in Fig. 1 F, whereas the number of TCR-MCs per cell was counted automatically using IMARIS software (BitPlane). First, we analyzed Ag dose dependency of micro–adhesion ring formation (Fig. 3 A and Video 1). The numbers of both micro–adhesion rings and TCR-MCs increased as cellular adhesion progressed, but the number of micro–adhesion rings at the higher doses of Ag (1 and 10 µM) decreased rapidly. In contrast, with lower doses (10 and 100 nM), micro–adhesion rings were generated and sustained for a longer period. Interestingly, whereas the number of TCR-MCs at 10 nM Ag was noticeably reduced, the number of micro–adhesion rings at 10 nM Ag was similar to that of 100 nM Ag. Because the adhesion area was also affected by the strength of TCR stimulation, we evaluated the densities of TCR-MCs and micro–adhesion rings upon stimulation with various Ag doses. The numbers of TCR-MCs or micro–adhesion rings at the peaks were divided by area (square micrometers) and plotted as the density (Fig. 3 B). Interestingly, the densities of TCR-MCs increased as the Ag dose increased, whereas those of micro–adhesion rings were constant except for those with no Ag. Without Ag, 17.4% of added cells showed transient adhesion, but they moved quickly on the bilayer. Consequently, only 1.6% of cells stopped and made a small number of TCR-MCs, which could represent preexisting MCs formed in the absence of Ag as reported previously (Fig. 3, A and C; Crites et al., 2014). Although we found that some of the preexisting MCs were surrounded by micro–adhesion rings (Fig. 3 A), they failed to recruit phosphorylated SLP76 (Fig. 3 D), suggesting that the micro–adhesion rings might not support delivering activation signals in the preexisting MCs. Considering that 10 nM is the minimum Ag dose for T cell stimulation based on the capacity to induce sustained calcium responses and produce IL-2 under this experimental condition (Fig. 3 E), these data suggest that the micro–adhesion ring plays a critical role in supporting T cell activation, particularly upon weak stimulation with low doses of Ag.

We next analyzed the Ag dose dependency in the clustering of ZAP70, LAT, and SLP-76 at TCR-MCs. With 1 µM Ag, all of the ZAP70-GFP, LAT-GFP, and SLP76-GFP formed clear clusters at the TCR-MCs and moved toward the center of the interface with TCR-MCs as previously reported (Fig. 3 F, left; and Video 2; Yokosuka et al., 2005). In contrast, upon stimulation with 10 nM Ag, LAT-GFP and SLP76-GFP formed significant numbers of distinct, clear clusters, but ZAP70-GFP showed only a small number of diffuse, unclear clusters (Fig. 3 F, right). The kinetics of the cluster number per cell was examined (Fig. 3 G). The number of ZAP70-GFP clusters by stimulation with 1 µM Ag initially increased and then decreased gradually and was comparable with that of TCR-MCs, but the number with 10 nM Ag was dramatically reduced. In the case of LAT-GFP, the number of clusters was quite similarly induced to that of ZAP70 upon stimulation with 1 µM Ag, but the number of LAT clusters by stimulation with 10 nM Ag was similar to those with 1 µM Ag. The number of SLP76 clusters by stimulation with 1 µM Ag also initially increased and then decreased more rapidly than those of TCR-MCs, and the number of SLP76 clusters by stimulation with 10 nM Ag was similar to those with 1 µM Ag. The kinetics of cluster intensities showed the difference between ZAP70, LAT, and SLP76-GFP clusters more clearly (Fig. 3 H). With 1 µM Ag, the intensities of TCR-MCs and ZAP70-GFP clusters gradually increased, whereas the SLP76-GFP clusters rapidly disappeared, and LAT-GFP clusters showed the intermediate pattern of ZAP70 and SLP76. TCR-MCs and ZAP70-GFP clusters upon stimulation with 10 nM Ag had much lower intensities than those with 1 µM Ag, whereas SLP76-GFP clusters showed similar intensities throughout the period, and LAT again showed the intermediate pattern of ZAP70 and SLP76. In this analysis, only the clusters generated within 1 min after attachment were analyzed because the clusters may have fused to each other later.

We detected a similar difference in the cluster formation between ZAP70 and SLP76 when using lower affinity Ag (a variant MCC K99A peptide) or T cells expressing low-affinity TCR (5c.c7-TCR tg T cells; unpublished data). These results demonstrate that ZAP70 clusters depend on the strength of TCR stimulation, whereas SLP76 clusters are relatively independent of TCR signal strength, and LAT clusters are formed in both a TCR signal–dependent and –independent fashion. TCR signal–independent LAT clustering and SLP76 clustering were similar to micro–adhesion ring formation. Consistently, almost all of the LAT clusters and SLP76 clusters were at the center of the micro–adhesion ring like small ISs upon 10 nM Ag stimulation (Fig. 3 I).

It has been shown that SLP76 clusters are induced not only by TCR stimulation, but also by the LFA-1–mediated outside-in signal (Baker et al., 2009). Therefore, we modulated the strength of the outside-in signal by changing the concentration of the ligand ICAM-1 on the planar bilayer (Fig. 4 A). The concentration of 200/µm² of ICAM-1 was used throughout the study because it is the equivalent level to that on the activated dendritic cells. When the ICAM-1 concentration was decreased to 20/µm², the number of micro–adhesion rings was reduced to half, whereas that of TCR-MCs was not changed. With 2/µm² of ICAM-1, the micro–adhesion rings were not induced at all, whereas TCR-MCs were still detected, suggesting that the generation of micro–adhesion rings strongly depends on the LFA-1 outside-in signal. The densities of micro–adhesion rings (the number per area) at the peak also decreased by reducing ICAM-1 concentration, whereas those of TCR-MCs were not significantly changed (Fig. 4 B). The number and density of SLP76 clusters by the outside-in signal showed similar results as the micro–adhesion rings (Fig. 4, A and B). Therefore, SLP76 clusters upon stimulation with low-dose Ag would be supported by the micro–adhesion ring. The intensity of TCR-MCs slightly increased with 2/µm² of ICAM-1, probably because of the excessive accumulation of TCRs through reduced integrin activation (Fig. 4, C and E). Because the majority of the cells did not adhere to the bilayer with 2/µm² of ICAM-1 and only a few cells were adherent and stopped, it may not represent a physiological condition (Fig. 4 D). Nevertheless, even under such conditions, SLP76-GFP clusters with 2/µm² of ICAM-1 had significantly lower intensity (Fig. 4 B). These data indicate that micro–adhesion rings depend on LFA-1–mediated outside-in signaling and that SLP76 clusters were predominantly supported by this structure.

It was shown that adhesion- and degranulation-promoting adapter protein (ADAP) regulates integrin-mediated cell adhesion from both inside-out and outside-in signals by stabilizing the recruitment of WASP to SLP76 for actin rearrangement (Peterson et al., 2001; Baker et al., 2009; Pauker et al., 2011). We confirmed that the SLP76 cluster in T cells from ADAP KO mice was severely reduced in intensity but was intact in number, which was probably supported by intact F-actin clusters (unpublished data). Therefore, we analyzed the contribution of F-actin on micro–adhesion rings.

F-actin distribution was analyzed by using the F-actin–binding peptide Lifeact. Lifeact-GFP was accumulated at TCR-MCs and in the peripheral lamellipodia area upon stimulation with 1 µM Ag, and they showed distinct, clear clusters, even with 10 nM Ag stimulation, which were colocalized with SLP76-GFP clusters and surrounded by Pxn-Halo rings (Fig. 5, A and B; and Video 3). Other F-actin–related molecules, WASP (Fig. 5, A and B) and the Arp2/3 complex (not depicted), were similarly detected at TCR-MCs upon stimulation with 10 nM Ag.

To confirm the importance of F-actin for the synapse-like structure, T cells were acutely treated with cytochalasin D (CytD), which promptly inhibits actin polymerization. CytD was added at the point of maximum spreading, and the time course of micro–adhesion rings and other cluster numbers are shown in Fig. 5 E. Inhibition of actin polymerization was confirmed by the rapid disappearance of Lifeact-GFP clusters after CytD addition (Fig. 5 E, bottom right). TCR-MCs and ZAP70-GFP clusters upon stimulation with 1 µM Ag were not changed by the addition of CytD and remained. These data were consistent with the previous study by Varma et al. (2006) that showed new generation of TCR-MCs was inhibited by F-actin inhibitor, whereas they were resistant to F-actin inhibitor after generation. The adhesion area was also comparable in CytD-treated T cells because CytD was added after the maximum expansion of the T cells (Fig. 5 G). In contrast, SLP76-GFP clusters and LAT-GFP clusters upon stimulation with 1 µM Ag decreased much more rapidly than the control by the addition of CytD. 10 nM Ag was used for Pyk2-GFP because they showed more prolonged micro–adhesion ring formation than with 1 µM Ag. Pyk2-GFP rings and SLP76-GFP clusters upon 10 nM Ag stimulation rapidly disappeared by CytD treatment similar to Lifeact-GFP. The images after the addition of CytD are shown in Fig. 5 F and Video 4.

Furthermore, pretreatment of T cells with Arp2/3 complex inhibitor CK666 (CK) was also used to inhibit F-actin. The T cells treated with 30 µM CK for 30 min dramatically reduced the number of Pyk2-GFP rings (Fig. 5 C), whereas the significant number of TCR-MCs remained, and the intensity of TCR-MCs decreased (Fig. 5 D).

These data clearly show that micro–adhesion rings, LAT clusters, and SLP76 clusters are dependent on F-actin, whereas TCR-MCs and ZAP70 clusters are relatively independent. The components of the synapse-like structure were clearly divided into two types: F-actin–dependent clustering (LAT, SLP76, micro–adhesion ring, etc.) and independent clustering (TCR, ZAP70, etc.).

Because the micro–adhesion ring is tightly associated and regulated by F-actin, we next examined the involvement of MyoII as an F-actin–related effector molecule for micro–adhesion ring function. T cells expressing Pxn-Halo or SLP76-Halo were loaded with tetramethylrhodamine (TMR) ligand and treated with 50 µM blebbistatin (Blebb), a MyoII inhibitor, for 3 h before micro–adhesion ring formation. Pretreatment with Blebb reduced Pxn-Halo rings and SLP76-Halo clusters, although it significantly retained the numbers and densities of TCR-MCs (Fig. 6 A), suggesting that MyoII activity is strongly related to the micro–adhesion ring formation. The kinetics of the cluster intensities showed that the TCR-MCs were slightly reduced and SLP76-Halo clusters were reduced and shortened (Fig. 6 B). Imaging analysis by using anti-MyoII Ab revealed that MyoII was colocalized with LFA-1 surrounding TCR-MCs (Fig. 6 C).

Therefore, these results suggest that MyoII and its activity are essential for the formation of synapse-like structures, especially for the recruitment of focal adhesion molecules. The impaired micro–adhesion ring would reduce the size of TCR-MCs and F-actin clusters and would result in defects in SLP76 clustering.

To investigate the importance of the micro–adhesion rings for TCR-MCs and T cell activation, RNA interference with shRNA for focal adhesion molecules was used. T cells were retrovirally transduced with shRNA for control (shCtrl), Pxn (shPxn), and Pyk2 (shPyk2). Down-regulation of Pxn and Pyk2 expression was confirmed by Western blotting (Fig. 7 A). The shCtrl-, shPxn-, and shPyk2-transduced cells on planar bilayers showed no significant difference in TCR-MC numbers upon stimulation with 1 µM Ag (Fig. 7 B). However, the cells cotransduced with shRNAs for Pxn and Pyk2 (double knockdown [dKD] cells) showed an increased number of TCR-MCs and impaired cSMAC formation (Fig. 7, B and C; and Video 5). dKD cells became more mobile on the planar membrane by stimulation with 1 µM Ag than shCtrl cells (Fig. 7 D). This unstable contact of dKD cells would be the reason for the impaired cSMAC formation. When T cells were stimulated with 10 nM Ag, the numbers of TCR-MCs in shPxn and dKD cells were dramatically reduced compared with that in shCtrl cells (Fig. 7 B and Video 5). The reduction in numbers appeared to depend on the size of the adhesion area because the density of clusters was not changed, suggesting that the effect might be based on dysregulation of focal adhesion (Fig. 7 E). Indeed, as expected, in the dKD cells, the number and size of each micro–adhesion ring detected by vinculin staining were decreased (Fig. 7 F).

Analyses of individual clusters revealed that the intensities of TCR-MCs in the dKD cells significantly decreased with both 1 µM and 10 nM Ag (Fig. 7 G). The Lifeact-GFP intensity in the dKD cells was also significantly decreased (Fig. 7 H). The phosphorylation status was analyzed at TCR-MCs by Ab staining. The numbers and intensities of the clusters of phospho-CD3ζ and phospho-SLP76 were decreased in the dKD cells (Fig. 7 I). These data indicate that the development of individual TCR-MCs and the signaling through TCR-MCs including phosphorylation and recruitment of SLP76 are supported by the micro–adhesion ring through regulation of cell adhesion.

To analyze the function of the synapse-like structure, dKD cells were stimulated with Ag peptide/APC, and T cell responses were analyzed. Upon interaction with APC, the frequency of cell adhesion and cSMAC formation was reduced in dKD cells (Fig. 8 A and Video 6). Consequently, flow cytometric analysis showed that phosphorylation of SLP76 and extracellular signal–regulated kinase (Erk [pErk]) upon stimulation with Ag peptide/APC was reduced in dKD cells (Fig. 8 B). IL-2 production upon stimulation with Ag peptide/APC was significantly reduced in dKD cells, whereas there was no difference even in dKD cells when stimulated by immobilized anti-CD3/CD28 Abs. Unlike molecular assembly and phosphorylation as early signaling, because IL-2 production is a late response, the reduction might be affected by reduced Pyk2/Pxn during a long period in addition to defective micro–adhesion rings (Fig. 8 C).

To determine the functional importance of the synapse-like structure further, we tried to dissect TCR and LFA-1 stimulation by using lipid bilayer–coated glass beads. We used three types of lipid bilayer–coated beads: only I-E^k–coated beads (E^k), both I-E^k– and ICAM-1–coated beads ([E^k + ICAM]), and a mixture of I-E^k beads and ICAM-1 beads ([E^k] + [ICAM]; Fig. 8 D, top). shRNA-treated T cells were stimulated by these glass beads with 1 µM Ag peptides, and the level of pErk as the representative activation signal was analyzed (Fig. 8 D). shCtrl cells stimulated with [E^k + ICAM] showed an increased level of pErk compared with those stimulated with E^k, demonstrating that ICAM-1 enhances pErk. In contrast, in shCtrl cells stimulated with [E^k] + [ICAM] separately, the level of pErk was almost comparable with those stimulated with E^k alone, suggesting that the ICAM-1 stimulation apart from I-E^k had little effect on T cell activation. The significance of the close proximity between I-E^k and ICAM-1 for T cell activation is consistent with and supports our idea that the synapse-like structure micro–adhesion ring surrounding TCR-MCs is critical for initial TCR signals. In contrast, the pErk induction upon stimulation with [E^k + ICAM] was much lower in dKD cells than that in shCtrl cells and was not significantly changed from that with [E^k] + [ICAM] (Fig. 8 D). The proportion of pErk-positive cells was counted, and the statistical results were plotted in Fig. 8 E (left). The same results were obtained in nontransfected CD4 T cells (Fig. 8 E, right). These results suggest that dKD cells could not be fully activated by [E^k + ICAM], probably because of the lack of close proximity of TCR and LFA-1 by the defect in micro–adhesion ring formation around TCR-MCs.

In this study, we describe the discovery of a specific structure of micro–adhesion rings just around TCR-MCs, which is required for early T cell activation (Fig. 8 F). The micro–adhesion ring is composed of integrin, focal adhesion molecules, and MyoII and surrounds each TCR-MC. Considering that the mature IS is composed of centralized TCRs and surrounding LFA-1, this structure resembles the IS in microscale, which we may call a microsynapse. The formation of the synapse-like structure is transient during the early stage of T cell activation and depends on LFA-1–mediated outside-in signals, F-actin, and MyoII activity. The micro–adhesion ring plays a critical role in facilitating adhesion, TCR clustering, and F-actin cluster formation to recruit LAT and SLP76 and generating TCR signals. Such function of the synapse-like structure to amplify and support TCR activation was particularly obvious upon weak stimulation with low doses of Ag.

It has been previously reported that LFA-1 induces dot clusters separately from TCR-MCs during the initial phase of T cell activation (Kaizuka et al., 2007). In the present study, we showed that LFA-1 was transiently accumulated to form a ring structure surrounding TCR-MCs. The ring structure was more clearly observed by Pxn-GFP or Pyk2-GFP than by CD11a-GFP (unpublished data), probably because of the abundant expression of CD11a-GFP on the cell surface. Indeed, more LFA-1 rings were detected by anti–LFA-1 staining than CD11a-GFP overexpression. Therefore, the number of the micro–adhesion rings identified by the transfected Pxn-GFP or Pyk2-GFP might be an underestimate of the total number.

The micro–adhesion ring has some similarities to the podosome/invadosome. The podosome is a cylindrical structure formed in a vertical direction on the ventral cell surface to support invasion of cancer cells or dendritic cells (Murphy and Courtneidge, 2011). The similarities include molecular components, size, dependency on receptor stimulation, outside-in signaling, MyoII activity, and structure with an F-actin core surrounded by adhesion molecules. (Collin et al., 2008; Albiges-Rizo et al., 2009; van den Dries et al., 2013). There has been a study describing a podosome-like protrusion at the interface between T cells and epithelial cells or APCs (Sage et al., 2012). This structure contained TCR, F-actin, and ZAP70, and the authors suggested its role in invading the carbohydrate forest on APC to find the peptide–MHC complex on the cell surface. Because the micro–adhesion rings are formed transiently during the early stage of T cell activation and are involved in the initial cell adhesion and activation, they may be similar to the podosome-like structure. Although we failed to detect such a small protrusion in T cells on the planar bilayer system using electron microscopy in our previous study (Hashimoto-Tane et al., 2011), the structural characterization of the observed synapse-like structure remains to be analyzed.

Because there is no means of specific disruption of the micro–adhesion ring to help determine its function, we analyzed T cell function after down-regulating two components of the micro–adhesion ring, Pxn and Pyk2. Consistent with a previous study of Pyk2 KO mice (Beinke et al., 2010), shPyk2 cells showed no structural and functional impairment. In contrast, the Pxn KO mice resulted in early lethality. It may have a role in growth and/or survival of cells, which is consistent with our data that a considerable proportion of shPxn cells failed to grow. Whereas the shPxn T cells had no detectable defects in adhesion or IS formation upon strong stimulation with 1 µM Ag, they showed some defects in cell adhesion and formation of TCR-MCs upon weak stimulation with 10 nM Ag. In contrast, the dKD cells exhibited severely impaired regulation of adhesion, TCR-MC formation, phosphorylation of signaling molecules, and cytokine production. Therefore, it is likely that the function of the micro–adhesion ring primarily depends on Pxn rather than Pyk2, and synergistic effects were observed by double deficiency of both Pxn and Pyk2 for the early stage of T cell activation.

The reduced size of TCR-MCs and impaired TCR proximal signaling were caused by the alteration of micro–adhesion rings. In addition, the defect in cell adhesion in dKD cells upon weak stimulation was also explained by impaired micro–adhesion ring formation because usually adhesion starts from the point of the initial TCR-MC and the generation and maturation of TCR-MCs depends on adhesion rings. In contrast, the increased adhesion area in dKD cells upon strong stimulation may reflect the effects from the general function of focal adhesions; stable adhesion in sessile cells and active cells move by recycling in moving cells (Webb et al., 2004). The dKD cells upon strong stimulation could induce adhesion but also failed to stop cells with narrowing down the adhesion area. The reduced IL-2 production could be mainly caused by the defective initial activation signal but may also be affected by the lack of Pxn or Pyk2 later during the 24-h culture period. The reduced sensitivity of T cell activation in the dKD cells was similarly observed in LFA-1–deficient T cells (Wang et al., 2008), which is consistent with impaired LFA-1 signaling in the dKD cells.

The colocalization of Pxn and Pyk2 suggested their cooperation at the micro–adhesion ring. According to previous studies (Rodríguez-Fernández et al., 1999; Herreros et al., 2000), it is possible that the cooperation of Pxn and Pyk2 at the MTOC regulates T cell polarization and activation. Indeed, we confirmed reduced efficiency of MTOC translocation in the dKD cells (unpublished data). However, the defects in MTOC translocation could also be explained by the defects in the initial T cell activation. Because the defective processes such as TCR-MC generation and phosphorylation of SLP76 clusters were very early membrane-proximal events, we suggest that the appropriate spatiotemporal generation of micro–adhesion rings with Pxn and Pyk2 is critical for inducing integrin outside-in signals required for T cell activation.

The discovery of the TCR-MC as the responsible structure and the site for inducing initial TCR activation signals revealed that TCR, ZAP-70, LAT, and SLP76 formed a TCR-MC, and their assembly and dynamics upon Ag stimulation were thought to be stoichiometric. However, in the present study, we showed that although the dynamics of TCR and ZAP-70 were almost identical, LAT and SLP76 behaved differently in terms of the kinetics of cluster numbers and intensities and the effects of an actin inhibitor, particularly upon weak TCR stimulation. Interestingly, the transient recruitment of SLP76 to TCRs in the very early phase of activation had also been shown in recent proteome analysis (Roncagalli et al., 2014). Whereas cluster formation of TCRs and ZAP70 was dependent on the strength of the activation signal, LAT and SLP76 clusters were formed relatively independently of signal strength. The observed difference in the dynamics of the LAT and SLP76 clustering from TCRs and ZAP70 was caused by its dependency on the formation of the micro–adhesion ring through LFA-1–mediated outside-in signals and F-actin. Recently, WASP-mediated F-actin–dependent clustering of phospho–PLC-γ was reported (Kumari et al., 2015). The F-actin supporting phospho–PLC-γ as actin foci is similar to the F-actin localized at the core of the synapse-like structure that supports SLP76 clusters and the micro–adhesion ring. Because the characteristics of the phospho–PLC-γ clusters were mostly consistent with LFA-1–mediated SLP76 clusters, it is likely that the PLC-γ clusters would also be under the regulation of the micro–adhesion ring.

In this study, we found the synapse-like structure micro–adhesion ring surrounding TCR-MCs and showed its importance in T cell activation. The adhesion-mediated positive regulation of TCR activation signals was clearly visualized. Such regulation of adhesion and the cytoskeleton will become more important in understanding the signaling and function of T cells.

AND-TCR tg mice were provided by S.M. Hedrick (University of California, San Diego, San Diego, CA) by courtesy of R.N. Germain (National Institutes of Health), and 5c.c7 tg mice were provided by M.M. Davis (Stanford University, Stanford, CA). ADAP-deficient mice were provided by G. Koretzky (Weill Cornell Medical College, New York, NY). C57BL/6 and B10.BR mice were purchased from CLEA Japan, Inc. and Japan SLC Inc., respectively. All mice were maintained in our animal facility in specific pathogen-free conditions and treated in accordance with the ethical guidance of the RIKEN Yokohama Institute.

T cells from the spleen and lymph node of AND-tg mice were isolated and purified by using CD4 (L3T4) MicroBeads (Miltenyi Biotec). The purified T cells were stimulated with irradiated spleen cells from B10.BR mice pulsed with 3 µM MCC mutant (MCC(88-103) K99A) peptide (MCC(K99A)) in RPMI 1640 (Sigma-Aldrich) with 10% FBS and 1% penicillin–streptomycin (Invitrogen). The next day, the genes of interest were transduced using a retrovirus-mediated gene transfer system. On day 4, the infected cells were sorted using a flow cytometer (FACS Aria; BD). Fluorescent imaging experiments were performed on the fifth or sixth day of culture. For shRNA treatment, the cells were restimulated with irradiated spleen and MCC(K99A) on the fifth or sixth day, transduced again with virus supernatants on the sixth or seventh day, and subjected to imaging and functional analysis on the 10th to 12th day of culture.

All the expression vectors were made by PCR subcloning. The retrovirus vector pMXs (provided by T. Kitamura, Tokyo University, Tokyo, Japan) were used for expression of fluorescent-labeled molecules. ZAP70, SLP76, ADAP, WASP, p16, Arp3, CapZ, Pyk2, Pxn, and CD11a were cloned from complementary DNA of activated AND-tg T cells. GFP was derived from plasmid enhanced GFP–N1 (BD), HaloTag from HaloTag pHT2 vector (Promega), and Lifeact from Lifeact-TagRFP2 pCMV (ibidi). The pMSCV-LTRmiR30-PIG(LMP) retrovirus vector (GE Healthcare) was used for shRNA expression. We used several different markers to detect doubly transduced cells. In addition to GFP, CD8 without the cytoplasmic region was used for shPxn or shCtrl, and rat CD2 without the cytoplasmic region was used for shPyk2 by inserting them after the internal ribosomal entry site. The sequences used were shPxn, 5′-CCCCTGGTGAAAGAGAAGCCAA-3′; shPyk2, 5′-AAGAAGTAGTTCTTAACCGCAT-3′; and shCtrl, 5′-GCACGGGGAAGGAGCGGGAA-3′. To produce pseudovirus, the recombinant plasmids were transfected into Phoenix packaging cells (G. Nolan, Stanford University, Stanford, CA) using Lipofectamine 2000 (Invitrogen).

For live imaging of TCR, the Fab fragment of anti–TCR-β Ab (H57) was labeled with Dylight549. anti-Pxn Ab (BD610051) and anti-Pyk2 Ab (D-1; sc-365201) were used for Western blotting. For immunohistochemistry, anti–CD3-ε Ab labeled with Alexa Flour 647 (BD557889), anti-vinculin Ab (F7053; Sigma-Aldrich), anti–LFA-1 Ab (H155-78; BioLegend) labeled with Dylight 549, and anti-MyoII Ab labeled with FITC (orb4013) were used. For FACS analysis, anti–phospho-Erk Ab (Phosflow 812593; BD) and anti–phospho-SLP76 Ab labeled with APCs (Phosflow 558438; BD) were used.

Glycophosphatidylinositol (GPI)-anchored mouse MHC class II I-E^k and ICAM-1 were purified as previously described (Grakoui et al., 1999). The purified I-E^k and ICAM-1 incorporated in dioleoylphosphatidylcholine liposomes (Avanti Polar Lipids, Inc.) were applied to cover glasses (thickness of 0.12–0.17 mm; Matsunami) or glass beads (4.86-µm sphere glass) to make the lipid bilayer containing 200 molecules of I-E^k and ICAM-1/µm² and incubated with Ag peptides (Yokosuka et al., 2005).

In this system, when 10 µM MCC(88-103) peptide was incubated at 37°C for 12 h, 8.9 ± 5.1% of total I-E^k was detected as peptide binding I-E^k (I-E^k + MCC; mean from three experiments) by staining with anti–I-E^k + MCC mAb D4 (Grakoui et al., 1999). Therefore, the concentration of I-E^k + MCC was 17.8/µm² for 10 µM. When we considered the general formula for affinity evaluation (K_d = [I-E^k]_free × [MCC]_free/[I-E^k + MCC]) and the published K_d value of 0.54 µM, which was measured by using soluble I-E^k and MCC (Reay et al., 1992), the I-E^k + MCC concentrations for 1 µM, 100 nM, and 10 nM MCC were estimated to be 11.3, 2.94, and 0.34 µm², respectively.

Hepes-buffered saline containing 1% BSA, 1 mM MgCl₂, and 1 mM CaCl₂ was used for the assay. For immunohistochemistry, cells were placed on a planar bilayer for the indicated periods and then fixed with 2% paraformaldehyde at 37°C for 10 min. For FACS analysis, cells were spun down with the glass beads and incubated for 10–15 min and fixed with 2% paraformaldehyde at 37°C for 10 min. The cells were permeabilized in 0.05% digitonin or 0.1% Triton X-100 and stained with Abs in PBS containing 5% BSA. For calcium measurement, cells were incubated with 5 µM Fluo-4 AM and 10 µM Fura-red AM in 0.5% BSA containing PBS for 20 min at room temperature.

A TIRF system (Olympus) was configured on a microscope (IX81). TIRF objective lenses (UAPON 100×, NA 1.49, oil; Olympus), dual excitation lasers, diode-pumped solid-state 488-nm and 561-nm laser systems (85BCD020 and 85YCA020; Melles Griot), and a charge-coupled device camera (ORCA flash4.0; Hamamatsu Photonics) were used. Dual-color videos were taken every 5 s for 10 min using Metamorph software (Molecular Devices). The images of the interaction between T cells and activated B cells that had been stimulated with LPS for 24 h were taken every 10 s using the IX81 microscope with an oil objective lens (Plan APO N 60×, NA 1.45; Olympus) under illumination with a Hg lamp. The intracellular calcium concentrations were also measured by using the IX81 microscope with Hg lamp illumination and an oil objective lens (UPlanAPO 40×, NA 1.00; Olympus). A microscope (TCS SP5; Leica Biosystems) with an oil immersion objective (HCX PL APO 100×, NA 1.44; Leica Biosystems) was used for fixed and stained samples. For detecting micro–adhesion rings formed in the cell–cell interaction, a hybrid detector (HyD; Leica Biosystems) was used to improve image quality.

IMARIS software (BitPlane) was used for counting and evaluation of fluorescent clusters. In this study, all the images were normalized using Fiji as follows: subtracted numerically until the mean background count became 10 and multiplied to adjust the count in the noncluster area of the cell to be 100. The criteria for a cluster was that the diameter was <5 µm, intensities at the center were >105–120 counts, and the summation of the fluorescence was >1,600–2,400 counts. The parameters were adjusted depending on the molecules being analyzed. The movements of the clusters were tracked automatically with an algorithm of Brownian movement, and the maximum gap distance in one frame was set as 5 µm.

Video 1 shows TCR-MCs transiently bordered by Pyk2 in the initial phase of activation. Video 2 shows different behaviors of ZAP70 and SLP76 clusters upon weak stimulation. Video 3 shows F-actin transiently accumulated at TCR-MCs in the initial phase of activation. Video 4 shows disappearance of SLP76 clusters and remaining TCR-MCs and ZAP70 clusters by CytD treatment. Video 5 shows the unstable contacts and impaired IS formation of dKD cells on the planer bilayer. Video 6 shows the unstable contacts and impaired IS formation of dKD cells upon interaction with APCs.

We thank K. Shiroguchi for helpful discussion, M. Unno, W. Kobayashi, and N. Suzuki for technical support, and M. Yoshioka and H. Yamaguchi for secretarial assistance.

This work was supported by the Japan Society for the Promotion of Science KAKENHI with Grants-in-Aid for Scientific Research to T. Saito ([S] JP24229004) and A. Hashimoto-Tane ([C] JP10415226) and partly by the Takeda Science Foundation (F1206042), Novartis Foundation, and Hayashi Memorial Foundation for Female Natural Scientists.

The authors declare no competing financial interests.

Author contributions: A. Hashimoto-Tane and T. Saito designed the studies. A. Hashimoto-Tane, M. Sakuma, and H. Ike performed experiments. T. Yokosuka provided advice and reagents. Y. Kimura and O. Ohara helped with molecular cloning. A. Hashimoto-Tane and T. Saito wrote the manuscript.